%% Grandfathering Project
%% Main regression first-stage table
%% Compiled on 2024-07-01 19:11:23 using <3_regs.R>
%% Written by Jack Gregory


\begin{table}[ht]
\centering
	\caption{Main regression first-stage results}
	\label{tbl:reg_main_fs}

\scriptsize

\begin{tabular}{@{\extracolsep{2.5ex}}l*{2}{D{.}{.}{-2}}@{\extracolsep{2.5ex}}}
\\[-1.8ex] \hline
\hline \\[-1.8ex]
 &	\multicolumn{1}{c}{(5)} &	\multicolumn{1}{c}{(7)}	\\\\ [-1.8ex]
 &	\multicolumn{1}{c}{\textit{OLS}} &	\multicolumn{1}{c}{\textit{OLS}}	\\\\ [-1.8ex]
 &	\multicolumn{1}{c}{\textit{IOU+}} &	\multicolumn{1}{c}{\textit{All}}	\\
\hline \\[-1.8ex]
Sulfur distance &	1.86^{***} &	1.18^{***} \\
 &	(5.94) &	(5.41) \\ [1.0ex]
GF &	-0.42^{*} &	-0.22 \\
 &	(-1.92) &	(-1.24) \\ [1.0ex]
size &	-1.37^{***} &	-0.81^{***} \\
 &	(-4.91) &	(-2.95) \\ [1.0ex]
GF $\times$ size &	1.17^{***} &	0.87^{***} \\
 &	(3.81) &	(2.95) \\ [1.0ex]
NAAQS &	-0.16 &	-0.00 \\
 &	(-0.65) &	(-0.02) \\ [1.0ex]
GF $\times$ NAAQS &	-0.02 &	-0.09 \\
 &	(-0.09) &	(-0.44) \\ [1.0ex]
MMBTU &	-0.01 &	-0.04 \\
 &	(-0.58) &	(-1.57) \\ [1.0ex]
GF $\times$ MMBTU &	0.04 &	-0.01 \\
 &	(1.25) &	(-0.34) \\ [1.0ex]
\hline \\[-1.8ex]
Observations &	\multicolumn{1}{c}{11,260} &	\multicolumn{1}{c}{17,253} \\
R$^{2}$ &	\multicolumn{1}{c}{0.607} &	\multicolumn{1}{c}{0.484} \\
\hline
\hline \\[-1.8ex]
\multicolumn{3}{p{9cm}}{\textit{Notes:} This table presents the first-stage results for our main regressions, corresponding to specifications (5) and (7) in Table 1. The three channels each produce similar results, so we report those for utilization only. The dependent variable is the SO$_2$ content of combusted coal. All specifications are estimated using OLS and data from 1995-2018. The unit of observation is boiler-year. The sample is restricted to those boilers with inservice years between 1950 and 2006, inclusive. The $IOU+$ column includes independently-owned utilities as well as commercial and industrial boilers; while, the $All$ column uses all types of available boilers. All regressions above include boiler and market controls as well as year, state, utility type and manufacture type fixed effects.  Boiler-level clustered standard errors used, with *** p$<$0.01,  ** p$<$0.05,  * p$<$0.10 and \textit{t}-statistics in parentheses.} \\
\end{tabular}
\end{table}
